Modeling the learner's cognitive state transitions using fuzzy logic techniques for adaptive learning
KeywordsΕξ αποστάσεως εκπαίδευση ; Ευφυές διδακτικό σύστημα ; Ασαφείς Γνωστικοί Χάρτες ; Fuzzy Cognitive Maps (FCMs) ; Intelligent tutoring systems
In this Ph.D. thesis a novel approach of web-based education that performs individualized instruction, adapting the delivery of the knowledge domain to the individual learner’s learning needs and pace, is presented. It includes fuzzy logic techniques to represent the learner’s knowledge and cognitive state. The presented Intelligent Tutoring System includes a rule-based fuzzy logic mechanism for providing personalized tutoring to each learner and an innovative module, which is responsible for tracking cognitive state transitions of learners with respect to their progress or non-progress. The presented approach models either how learning progresses or how the student’s knowledge can be decreased. In particular, it performs user modeling by dynamically identifying and updating the student’s knowledge level for all the concepts of the domain knowledge. Its operation is based on Fuzzy Cognitive Maps (FCMs). They are used to represent the dependencies among the domain concepts. The presented student model uses fuzzy sets to represent the student’s knowledge level as a subset of the domain knowledge. Thus, it combines fuzzy theory with the overlay model. Moreover, it employs a novel inference mechanism that dynamically updates user stereotypes using fuzzy sets. This mechanism is triggered after any change of the student’s knowledge level of a domain concept. Then, it updates the student’s knowledge level of the concepts, which are related with the concept that the student has learnt or forgotten. The transition of a learner from one stereotype to another reveals her/his learning state each time. In particular, it reveals if a student learns or not, if s/he forgets and reasons these states. The presented novel approach was fully implemented and evaluated. Particularly, an original integrated environment for personalized e-training in programming and the programming language C, which is called ELaC, was developed. The specific knowledge domain was chosen due the fact that in the domain of computer programming there are many different programming languages and learners have a variety of different backgrounds and characteristics. Therefore, it is suitable for the implementation and evaluation of the thesis’ issue. ELaC incorporates the presented fuzzy student modeling approach. Thereby, recognizes when a new domain concept is completely unknown to the learner, or when it is partly known due to the learner having previous related knowledge. Furthermore, it recognizes when a previously known domain concept has been completely or partly forgotten by the learner. This system was used by the students of a postgraduate program in the field of Informatics in the University of Piraeus, Greece, in order to learn how to program in the programming language C. For the evaluation of the fuzzy student model approach, two well-known and common-used evaluation methods were chosen: the Kirkpatrick’s model and the layered evaluation method. The results of the evaluation were very encouraging. They demonstrated that the system models the student’s cognitive state and adapts dynamically to her/his individual needs by scheduling the sequence of lessons instantly, allowing her/him to complete the e-training course at her/his own pace and according to her/his ability. It has to be noted that the presented novel combination of overlay model and stereotypes with fuzzy sets is significant as the students’ level of knowledge is represented in a more realistic way by automatically modeling the learning or forgetting process of a student with respect to the FCMs. The application of this approach is not limited to adaptive instruction, but it can also be used in other systems with changeable user states, such as e-shops, where consumers’ preferences change over the time and affect one another. Therefore, the particular approach constitutes a novel generic fuzzy tool, which offers dynamic adaptation to users’ needs and preferences of adaptive systems.